Cycle-consistent generative adversarial neural networks based low quality fingerprint enhancement

Abstract

Distortions such as dryness, wetness, blurriness, physical damages and presence of dots in fingerprints are a detriment to a good analysis of them. Even though fingerprint image enhancement is possible through physical solutions such as removing excess grace on the fingerprint or recapturing the fingerprint after some time, these solutions are usually not user-friendly and time consuming. In some cases, the enhancements may not be possible if the cause of the distortion is permanent. In this paper, we are proposing an unpaired image-to-image translation using cycle-consistent adversarial networks for translating images from distorted domain to undistorted domain, namely, dry to not-dry, wet to not-wet, dotted to not-dotted, damaged to not-damaged, blurred to not-blurred. We use a database of low quality fingerprint images containing 11541 samples with dryness, wetness, blurriness, damages and dotted distortions. The database has been prepared by real data from VISA application centres and have been provided for this research by GEYCE Biometrics. For the evaluation of the proposed enhancement technique, we use VGG16 based convolutional neural network to assess the percentage of enhanced fingerprint images which are labelled correctly as undistorted. The proposed quality enhancement technique has achieved the maximum quality improvement for wetness fingerprints in which 94% of the enhanced wet fingerprints were detected as undistorted.

On restoration of degraded fingerprints

“…Joshi et al [53] [54] propose a generative adversarial network (GAN) based fingerprint enhancement algorithm (FP-E-GAN) and show its superior performance. Karabulut et al [55] propose a cycle-consistent GAN for unpaired translation from distorted domain to non-distorted domain. Recently, Joshi et al [56] propose introduce data uncertainty in fingerprint enhancement models and show that data uncertainty guided GAN (DU-GAN) outperforms state-of-the-art fingerprint enhancement models.…”

“…This section compares the performance of the proposed CR-GAN with the state-ofthe-art generative adversarial network based fingerprint enhancement models Cycle-GAN [55] and DU-GAN [56]. Figure 15 (a) and Table 13 compare the average NFIQ fingerprint quality sore obtained on enhanced images generated using Cycle-GAN, DU-GAN and the proposed CR-GAN.…”

Deep Learning Algorithms based Fingerprint Authentication: Systematic Literature Review

“…The proposed method is evaluated on the CASIA Fingerprint dataset produced the best results using SIFT and ICA for fingerprint recognition with accuracy of 96.69, specificity f 96.77, precision of 94.88, recall 95.94 and F1-score of 95.4. A study to improve the quality of fingerprint digitally was proposed by [96]. The paper proposed a deep network to avoid the time consumption during physical recapturing of fingerprint.…”

“…Few studies on GAN demonstrate the efficacy of DL in generating fingerprint images. Different application tasks of the proposed DL-based methods have been identified such as fingerprint classification [15, 67, 79-81, 97, 100, 101, 111, 112], fingerprint liveness detection [65,74,75,77,106,110], fingerprint recognition and authentication [25,26,76], overlapped fingerprint separation [86], double-identity fingerprint detection [87], fingerprint ROI segmentation [88,89], fingerprint alteration detection [94], fingerprint image enhancement [20,96,107,108], latent fingerprint segmentation [83], latent fingerprint recognition [85], latent fingerprint enhancement [84], fingerprint indexing [90,91], fingerprint pore matching [70,99], partial fingerprint matching [92,93], cancelable recognition system [95], fingerprint spoofing detection [6,109], contactless to contact-based and 3D partial fingerprint images matching [104,105], fingerprint minutiae extraction [71,102,103], fingerprint pore extraction [98], fingerprint generation, and presentation attack detection [113,114], fingerprint recovery scheme [115], and fingerprint l...…”

Prediction of sgRNA Off-Target Activity in CRISPR/Cas9 Gene Editing Using Graph Convolution Network

“…However, these traditional machine learning methods cannot take raw features from large, annotated datasets and use them to identify the patterns buried inside them. Deep learning algorithms is a powerful approach for learning complex patterns and has led to multiple performance breakthroughs in many research fields, including computer vision [ 32 , 33 , 34 , 35 , 36 , 37 ] and natural language processing [ 38 , 39 , 40 ]. However, very few prediction models have implemented the concept of deep learning into the sgRNA off-target propensity prediction problem.…”